In this piece of work using only three grid points, we propose two sets of numerical methods in a coupled manner for the solution of fourth-order ordinary differential equation uiv(x)=f(x,u(x),u′(x),u′′(x),u′′′(x)), a<x<b, subject to boundary conditions u(a)=A0, u′(a)=A1, u(b)=B0, and u′(b)=B1, where A0, A1, B0, and B1 are real constants. We do not require to discretize the boundary conditions. The derivative of the solution is obtained as a byproduct of the discretization procedure. We use block iterative method and tridiagonal solver to obtain the solution in both cases. Convergence analysis is discussed and numerical results are provided to show the accuracy and usefulness of the proposed methods.
1. Introduction
Consider the fourth-order boundary value problem
uiv(x)=f(x,u(x),u′(x),u′′(x),u′′′(x)),a<x<b,
subject to the prescribed natural boundary conditionsu(0)=A0,u′(0)=A1,u(1)=B0,u′(1)=B1,
or equivalently, for u′(x)=v(x),uiv(x)=f(x,u(x),v(x),u′′(x),v′′(x)),a<x<b,
subject to the natural boundary conditions
u(0)=A0,v(0)=A1,u(1)=B0,v(1)=B1,
where A0, A1, B0, and B1 are real constants and -∞<a≤x≤b<∞.
Fourth-order differential equations occur in a number of areas of applied mathematics, such as in beam theory, viscoelastic and inelastic flows, and electric circuits. Some of them describe certain phenomena related to the theory of elastic stability. A classical fourth-order equation arising in the beam-column theory is the following (see Timoshenko [1]):EId4udx4+Pd2udx2=q,
where u is the lateral deflection, q is the intensity of a distributed lateral load, P is the axial compressive force applied to the beam, and EI represents the flexural rigidity in the plane of bending. Various generalizations of the equation describing the deformation of an elastic beam with different types of two-point boundary conditions have been extensively studied via a broad range of methods.
The existence and uniqueness of solutions of boundary value problems are discussed in the papers and book of Agarwal and Krishnamoorthy, Agarwal and Akrivis (see [2–5]). Several authors have investigated solving fourth-order boundary value problem by some numerical techniques, which include the cubic spline method, Ritz method, finite difference method, multiderivative methods, and finite element methods (see [6–16]). In the 1980s, Usmani et al. (see [17–19]) worked on finite difference methods for solving [p(x)y′′]′′+q(x)y=r(x) and finite difference methods for computing eigenvalues of fourth-order linear boundary value problem. In 1984, Twizell and Tirmizi (see [20]) developed multi-derivative methods for linear fourth-order boundary value problems. In 1984, Agarwal and Chow (see [21]) developed iterative methods for a fourth-order boundary value problem. In 1991, O’Regan (see [13]) worked on the solvability of some fourth-(and higher) order singular boundary value problems. In 1994, Cabada (see [22]) developed the method of lower and upper solutions for fourth- and higher-order boundary value problems. In 2005 Franco et al. (see [23]) dealt with some fourth-order problems with nonlinear boundary conditions. In 2006, Noor and Mohyud-Din (see [12]) used the variational iteration method to solve fourth-order boundary value problems and further developed the homotopy perturbation method for solving fourth order boundary value problems. Some of these methods use transformation in order to reduce the equation into more simple equation or system of equations and some other methods give the solution in a series form which converges to the exact solution. Later, Han and Li (see [24]) worked on some fourth-order boundary value problems.
In this paper we present the finite difference methods for the solution of fourth-order boundary value problem. We discretize the interval [a,b] into N+1 subintervals each of width h=(b-a)/(N+1), where N is a positive integer. We seek the solution of (la) or (2a) at the grid points, xk=kh, k=1,2,…,N. Let uk and uk′ denote the approximate solutions, and let Uk=u(xk) and Uk′=u′(xk) be the exact solution values of u(x) and u′(x) at the grid point x=xk, respectively. Also, x0=a and xN+1=b.
Using the second-order central differences, we obtain a five-point difference formula for (1.1), which requires the use of fictitious points outside [a,b]. The accuracy of the numerical solution depends upon the boundary approximation used. The finite difference method discussed here is based only on three grid points for second-and fourth-order methods. Therefore, no fictitious points are required for incorporating the boundary conditions. Here, we use a combination of the value of the solution u(x) and its derivative u′(x) to derive the difference scheme using three grid points.
Since we need to solve a coupled system of equations at each mesh point, the block successive overrelaxation (BSOR) iterative method is used.
2. The Finite Difference Method
The method is described as follows: for k=1(1)N, letu̅k′′=(uk+1′-uk-1′)2h,u̅k′′′=(uk+1′-2uk′+uk-1′)h2,f̅k=f(xk,uk,uk′,u̅k′′,u̅k′′′).
Then, the difference method of order two for the given differential (1.1) is given by,-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=h46f̅k,
and the corresponding difference method for the derivative u′(x) is given by-3(uk+1-uk-1)+h(uk+1′+4uk′+uk-1′)=0.Also, let u̅k±1′′=(±3uk±1′∓4uk′±uk∓1′)2h,u̅̅k′′′=15(uk+1-uk-1)2h3-3(uk+1′+8uk′+uk-1′)2h2,u̅̅k±1′′=-(11uk±1-16uk+5uk∓1)2h2±(4uk±1′-uk∓1′)h,u̅̅k′′=2(uk+1-2uk+uk-1)h2-(uk+1′-uk-1′)2h,u̅k±1′′′=∓(27uk±1-48uk+21uk∓1)2h3+(15uk±1′-9uk∓1′)2h2,u̅̅k±1′′′=∓(99uk±1-48uk-51uk∓1)2h3+(39uk±1′+96uk′+15uk∓1′)2h2,
and setf̅k±1=f(xk±1,uk±1,uk±1′,u̅k±1′′,u̅k±1′′′),f̅̅k±1=f(xk±1,uk±1,uk±1′,u̅̅k±1′′,u̅̅k±1′′′),f̅̅k=f(xk,uk,uk′,u̅̅k′′,u̅̅k′′′).Then, the difference method of order four for the differential equation and the corresponding difference method for the derivative uk′ are given by-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=h490(f̅̅k+1+f̅̅k-1+13f̅̅k),-3(uk+1-uk-1)+h(uk+1′+4uk′+uk-1′)=h460(f̅k+1-f̅k-1).
Note that u0, u0′, uN+1 and uN+1′, are prescribed. It is convenient to express the above finite difference schemes in block tridiagonal matrix form. If the differential equation is linear, the resulting block tridiagonal linear system can be solved using the block Gauss-Seidel (BGS) iterative method. If the differential equation is nonlinear, the system can be solved using the Newton nonlinear block successive overrelaxation (NBSOR) method (see [25, 26]).
3. Derivation of the Difference Scheme
For the derivation of the method we follow the approaches given by Chawla [27] and Mohanty [10]. In this section we discuss the derivation of the difference methods and the block iterative methods.
At the grid point xk, the given differential equation can be written asukiv=f(xk,uk,uk′,uk′′,uk′′′)=fk,k=1(1)N.
Similarly,fk±1=f(xk±1,uk±1,uk±1′,uk±1′′,uk±1′′′),k=1(1)N.
Using the Taylor expansion about the grid point xk, we first obtain-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=h490(fk+1+fk-1+13fk)+T1,
where T1=O(h8).
Now, we need the O(h2) approximation for uk±1′′′. Letu̅k+1′′′=1h3[a10uk+a11uk+1+a12uk-1]+1h2[b10uk′+b11uk+1′+b12uk-1′].
Expanding each term on the right-hand side of (3.3) in the Taylor series about the point xk and equating the coefficients of hp (p=-3,-2,-1,0, and 1)to zero, we get(a10,a11,a12,b10,b11,b12)=(24,-272,-212,0,152,-92).
Thus, we obtainu̅k+1′′′=-12h3[27uk+1-48uk+21uk-1]+12h2[15uk+1′-9uk-1′]=uk+1′′′-2h25ukv+O(h3).
Similarly, we obtainu̅k-1′′′=12h3[27uk-1-48uk+21uk+1]+12h2[15uk-1′-9uk+1′]=uk-1′′′-2h25ukv-O(h3).
Further, from (2.3) we obtain:u̅k±1′′=uk±1′′-h23ukiv±O(h3).
Hence, we can verify that f̅k±1=fk±1+O(h2).
Next, we obtain the O(h4) approximation for uk′′. Letu̅̅k′′=1h2[a20uk+a21uk+1+a22uk-1]+1h[b20uk′+b21uk+1′+b22uk-1′].
Using the Taylor series expansion and equating the coefficients of hp (p=-2,-1,0,1,2, and 3) to zero, we get(a20,a21,a22,b20,b21,b22)=(-4,2,2,0,-12,12).
Therefore,
u̅̅k′′=2h2[uk+1-2uk+uk-1]-12h[uk+1′-uk-1′]=uk′′+O(h4).
Similarly,
u̅̅k′′′=152h3[uk+1-uk-1]-32h2[uk+1′+8uk′+uk-1′]=uk′′′+O(h4),u̅̅k±1′′=-12h2[11uk±1-16uk+5uk∓1]±1h[4uk±1′-uk∓1′]=uk±1′′∓h315ukv+O(h4).
Next, we obtain the O(h3) approximation for uk+1′′′. Letu̅̅k+1′′′=1h3[a30uk+a31uk+1+a32uk-1]+1h2[b30uk′+b31uk+1′+b32uk-1′].
Equating the coefficients of hp (p=-3,-2,-1,0,1, and 2) to zero, we get(a30,a31,a32,b30,b31,b32)=(24,-992,512,48,392,152).
Thus, we obtainu̅̅k+1′′′=-12h3[99uk+1-48uk-51uk-1]+12h2[39uk+1′+96uk′+15uk-1′]=uk+1′′′-h310ukvi+O(h4).
Similarly,u̅̅k-1′′′=12h3[99uk-1-48uk-51uk+1]+12h2[39uk-1′+96uk′+15uk+1′]=uk-1′′′+h310ukvi+O(h4).
Let
αk=∂f∂uk′′,βk=∂f∂uk′′′.
From (3.5a), (3.5b), (3.6), and (2.9a) it follows that f̅k±1 provides the O(h2) approximation for fk±1 andf̅k±1=fk±1-h23ukivαk-2h25ukvβk±O(h3).
Also, from (2.9b), (3.11), (3.14), and (3.15) we obtain the O(h3) approximation for fk±1:f̅̅k±1=fk±1∓h315ukvαk∓h310ukviβk+O(h4).
From (2.9c), (3.9), and (3.10) it follows that f̿k provides an O(h4) approximation for fkf̅̅k=fk+O(h4).
From (3.18) and (3.19) we getf̅̅k+1+f̅̅k-1+13f̅̅k=fk+1+fk-1+13fk+O(h4).
With the help of (3.2) and (3.20), from (2.10), we obtain that the local truncation error associated with the difference scheme (2.10) is of O(h8). Similarly, we can verify that the local truncation error associated with the difference scheme (2.11) is of O(h7).
Thus, we have the following result.
Theorem 3.1.
Consider the fourth-order nonlinear ordinary differential equation (1.1) along with the boundary conditions (1.2). Let u∈C8[a,b], and the function f is sufficiently differentiable with respect to its arguments. Then, the difference methods (2.10) and (2.11) with the approximations of u′′ and u′′′ listed in (2.3)–(2.8) are of O(h4).
If the differential (1.1) is linear, then the difference method (2.10) and (2.11) in the matrix form can be written as[A11A12A21A22][uu′]=[d1d2],
where A11, A12, A21, and A22 are the Nth-order tri-diagonal matrices and d1 and d2 are vectors consisting of right-hand side functions and some boundary conditions associated with the block system.
The block successive over relaxation (BSOR) method (See Mohanty and Evans [26, 28]) is given by
A11u(n+1)=ω[-A12(u′)(n)+d1]+(1-ω)A11u(n),n=0,1,2,…,A22(u′)(n+1)=ω[-A21u(n+1)+d2]+(1-ω)A22(u′)(n),n=0,1,2,…,
where ω is a parameter known as relaxation parameter. With ω=1, the BSOR method reduces to block Gauss-Seidel (BGS) method. If ω>1 or ω<1, we have overrelaxation or underrelaxation, respectively.
If f(x,u(x),v(x),u′(x),v′(x))is nonlinear, the difference equations (2.2a), (2.2b) or (2.10), (2.11) form a coupled nonlinear system. To solve the coupled nonlinear system we apply the Newton NBSOR method.
We first write the difference equations (2.2a), (2.2b) or (2.10), (2.11) asΦ(u,v)=0,Ψ(u,v)=0,
whereu=[u1u2⋮uN],v=[v1v2⋮vN],Φ(u,v)=[ϕ1(u,v)ϕ2(u,v)⋮ϕN(u,v)],Ψ(u,v)=[ψ1(u,v)ψ2(u,v)⋮ψN(u,v)].
LetJ=[T11T12T21T22]
be the Jacobian of Φ and Ψ, which is the 2Nth-order block tridiagonal matrix, where
T11=∂(ϕ1,ϕ2,…,ϕN)∂(u1,u2,…,uN)=[∂ϕ1∂u1∂ϕ1∂u2∂ϕ2∂u1∂ϕ2∂u2∂ϕ2∂u300∂ϕN∂uN-1∂ϕN∂uN],T12=∂(ϕ1,ϕ2,…,ϕN)∂(v1,v2,…,vN),T21=∂(ψ1,ψ2,…,ψN)∂(u1,u2,…,uN),T22=∂(ψ1,ψ2,…,ψN)∂(v1,v2,…,vN)
are the Nth-order tridiagonal matrices.
In the NBSOR method starting with any initial approximation (u(0),v(0)) of (u(k),v(k)), k=1(1)N, we defineu(n+1)=u(n)+Δu(n),n=0,1,2,…,v(n+1)=v(n)+Δv(n),n=0,1,2,…,
where Δu and Δv are intermediate values obtained by solving the matrix equation for NBSOR method given by[T11T12T21T22][ΔuΔv]=[-Φ-Ψ].
The above system can be solved for Δu(n) and Δv(n) by using the block SOR method (inner iterative method) as follows:T11Δu(n+1)=ω[-Φ(u(n),v(n))-T12Δv(n)]+(1-ω)T11Δu(n),n=0,1,2,…,T22Δv(n+1)=ω[-Ψ(u(n),v(n))-T21Δu(n+1)]+(1-ω)T22Δv(n),n=0,1,2,…,
where ω∈(0,2) is a relaxation parameter and n=0,1,2,…. The above system of equations can be solved by using the tridiagonal solver. In order for this method to converge it is sufficient that the initial approximation (u(0),v(0)) be close to the solution.
4. Convergence and Stability Analysis
Consider the model problem uiv=f(x), where f is a function of x only. Applying the fourth-order difference methods (2.10) and (2.11) to the above equation, we get(uk-1-2uk+uk+1)+h2(vk-1-vk+1)=-h4180(fk+1+fk-1+13fk),3h(uk-1-uk+1)+(vk-1+4vk+vk+1)=h360(fk+1-fk-1),
where we denote u′=v.
Let us denote by P=[1,0,1], L=[1,1,1], and M=[1,0,-1] the Nth-order tridiagonal matrices.
The system of (4.1) can be written in the block form as
[(L-3I)h2M3hM(L+3I)][uv]=[d1d2],
where u and v are two N-dimensional solution vectors and d1, d2 are vectors consisting of right-hand side functions and some boundary values associated with (4.1). The BSOR method for the scheme isu(n+1)=(1-ω)u(n)-ωh2(L-3I)-1M(v)(n)+ω(L-3I)-1d1,(v)(n+1)=(1-ω)(v)(n)-3ωh(L+3I)-1Mu(n+1)+ω(L+3I)-1d2,
The associated block SOR and block Jacobi iteration matrices are given byLω=[(1-ω)I-ωh2(L-3I)-1M-3ωh(L+3I)-1M(1-ω)I],B=[0-h2(L-3I)-1M-3h(L+3I)-1M0],
and λ and η are the eigenvalues associated with the corresponding matrices Lω and B, which are related by the equation(λ+ω-1)2=λω2η2.
Let [v1v2] be the eigenvector associated with the eigenvalue η so that[0-h2(L-3I)-1M-3h(L+3I)-1M0][v1v2]=[v1v2]η.
That is,-h2(L-3I)-1Mv2=ηv1,-3h(L+3I)-1Mv1=ηv2.
On eliminating v2, we obtain32(L-3I)-1M(L+3I)-1Mv1=η2v1.
The rate of convergence of the BSOR method is given by -log(ρ(Lω)).
The rate of convergence of the BSOR method is dependent on the eigenvalues of B (through relation (4.5)), which are given by (3τ/2)=η2, where τ are the eigenvalues of (L-3I)-1M(L+3I)-1M.
Hence, we can determine the optimal parameter as
ω0=21+1-(3/2)τ̅,
where τ̅=S((L-3I)-1M(L+3I)-1M).
Thus, we can determine the convergence factor
λ̅=ω0-1=1-1-(3/2)τ̅1+1-(3/2)τ̅.
For convergence, we must have |λ̅|<1 to give the range
0<τ̅<23
Hence, we get the convergence.
Thus, we have the following result.
Theorem 4.1.
The iteration method of the form (4.3) for the solution of uiv=f(x) converges if 0<τ̅<2/3, where τ̅=S((L-3I)-1M(L+3I)-1M), L=[1,1,1],andM=[1,0,-1] are the N×N tridiagonal matrices.
Now, we discuss the stability analysis.
An iterative method for (4.1) can be written as u(k+1)=12Pu(k)+h4Mv(k)+RHU,v(k+1)=-34hMu(k)-14Pv(k)+RHV,
where u(k), v(k) are solution vectors at the kth iteration and RHU, RHV are right-hand side vectors consisting of boundary and homogenous function values.
The above iterative method in matrix form can be written as [u(k+1)v(k+1)]=G[u(k)v(k)]+RH,
whereG=[12Ph4M-34hM-14P],RH=[RHURHV].
The eigenvalues of P and M are 2coskπ/(N+1) and 2icoskπ/(N+1), respectively, where k=1,2,…,N. The characteristic equation of the matrix G is given bydet[12P-ξINh4M-34hM-14P-ξIN]=0.
Thus the eigenvalues of G are given by
det[-14P-ξIN]×det[(12P-ξIN)+316M(-14P-ξIN)-1M]=0.
The proposed iterative method (4.13) is stable as long as the maximum absolute eigenvalues of the iteration matrix are less than or equal to one. It has been verified computationally that all eigenvalues of the system (4.16) are less than one. Hence, the iterative method (4.1) is stable.
5. Application to Singular Equation
Consider a singular fourth-order linear ordinary differential equation of the formΔ4u≡(d2dr2+1rddr)2u=f(r),0<r<1,
or equivalentlyuiv=b(r)u′′′+c(r)u′′+d(r)u′+f(r),0<r<1,
where
b(r)=-2/r,c(r)=1/r2,d(r)=-1/r3.
The above equation represents fourth-order ordinary differential equation in cylindrical polar coordinates.
The boundary conditions are given byu(0)=A0,u′(0)=A1,u(1)=B0,u′(1)=B1,
where A0, A1, B0, and B1 are constants.
Applying the difference scheme (2.2a), (2.2b) to the singular equation (5.2), we obtain a second-order difference method-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=h46(bku̅k′′′+cku̅k′′+dkuk′+fk),-3(uk+1-uk-1)+h(uk+1′+4uk′+uk-1′)=0,k=1(1)N.
Applying the fourth-order difference scheme (2.10) to the singular equation (5.2), we obtain-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=h490[(bk+1u̅̅k+1′′′+ck+1u̅̅k+1′′+dk+1uk+1′+fk+1)+(bk-1u̅̅k-1′′′+ck-1u̅̅k-1′′+dk-1uk-1′+fk-1)+(13bku̅̅k′′′+13cku̅̅k′′+13dkuk′+13fk)],k=1(1)N,
where
bk±1=b(rk±1),ck±1=c(rk±1),dk±1=d(rk±1),fk±1=f(rk±1).
Note that the scheme fails when the solution is to be determined at k=1. We overcome the difficulty by modifying the method in such a way that the solutions retain the order and accuracy even in the vicinity of the singularity r=0 (see [29]). We consider the following approximations:bk±1=bk±hbk′+h22!bk′′+O(±h3),ck±1=ck±hck′+h22!ck′′+O(±h3),dk±1=dk±hdk′+h22!dk′′+O(±h3),fk±1=fk±hfk′+h22!fk′′+O(±h3).
Using the approximation (5.8) in (5.6) and neglecting higher-order terms, we can rewrite (5.6) in compact operator form as-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=h290(-24bk′+18ck-4h2ck′′)δx2uk+h180(45bk-75h2bk′′-6h2ck′)(2μxδx)uk+h490(-45bkh2+75bk′′+6ck′+15dk+h2dk′′)uk′+h2180(15bk+27h2bk′′+6h2ck′+2h2dk)(2μxδx)uk′+h3180(24bk′-3ck+5h2ck′′+2h2dk′)(2μxδx)uk+h490(15fk+h2fk′′),k=1(1)N.
Similarly, using the difference scheme (2.11), a fourth-order approximation for the derivative u′ for the singular equation (5.2) in the compact form may be written as-3(uk+1-uk-1)+h(uk+1′+4uk′+uk-1′)=h60(-24bk)δx2uk+h260(-3bk′)(2μxδx)uk+h360(2ck+3bk′)δx2uk′+h260(12bk+h2(ck′+dk))(2μxδx)uk′+h360(6bk′+2h2dk′)uk′+h460(2hfk′),k=1(1)N,
where δruk=uk+1/2-uk-1/2, μruk=(uk+1/2+uk-1/2)/2.
Finite difference equations (5.9) and (5.10) along with the boundary conditions (5.4) gives a 2N×2N linear system of equations for the unknowns u1,u2,…,uN,u1′,u2′,…,uN′. The resulting block tri-diagonal system can be solved using the BGS method. The schemes (5.9) and (5.10) are free from the terms 1/(k±1), hence very easily solved for k=1(1)N in the region (0,1).
Consider the coupled nonlinear singular equationsuIV=a(r)[u′v′′+v′u′′]+f(r),0<r<1,vIV=-a(r)u′u′′+g(r),0<r<1,
where a(r)=1/r, with known boundary conditions u(0), v(0), u′(0), v′(0), u(1), v(1), u′(1), and v′(1). The coupled equations represent model equations of equilibrium for a load symmetrical about the centre (see [30]).
The second-order difference scheme for solving the system (5.11) is given by-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=h46[ak(uk′v̅k′′+vk′u̅k′′)+fk],-2(vk+1-2vk+vk-1)+h(vk+1′-vk-1′)=h46[-akuk′u̅k′′+gk],-3(uk+1-uk-1)+h(uk+1′+4uk′+uk-1′)=0,-3(vk+1-vk-1)+h(vk+1′+4vk′+vk-1′)=0,
where ak=a(rk), fk=f(rk), and gk=g(rk).
The fourth-order difference scheme for solving the system (5.11) for u, v, u′, and v′ is given by
-2(uk+1-2uk+uk-1)+h(uk+1′-uk-1′)=H1[15fk+h2fk′′]+H1a11[uk+1′{-12h2(11vk+1-16vk+5vk-1)+1h(4vk+1′-vk-1′)}+vk+1′{-12h2(11uk+1-16uk+5uk-1)+1h(4uk+1′-uk-1′)}]+H1a12[uk-1′{-12h2(11vk-1-16vk+5vk+1)-1h(4vk-1′-vk+1′)}+vk-1′{-12h2(11uk-1-16uk+5uk+1)-1h(4uk-1′-uk+1′)}]+H1a10[uk′{2h2(vk+1-2vk+vk-1)-12h(vk+1′-vk-1′)}+vk′{2h2(uk+1-2uk+uk-1)-12h(uk+1′-uk-1′)}],-2(vk+1-2vk+vk-1)+h(vk+1′-vk-1′)=H1[15gk+h2gk′′]-H1a11uk+1′[-12h2(11uk+1-16uk+5uk-1)+1h(4uk+1′-uk-1′)]-H1a12uk-1′[-12h2(11uk-1-16uk+5uk+1)-1h(4uk-1′-uk+1′)]-13H1a10uk′[2h2(uk+1-2uk+uk-1)-12h(uk+1′-uk-1′)],-3(uk+1-uk-1)+h(uk+1′+4uk′+uk-1′)=H2(2hfk′)+H2a11[uk+1′{3vk+1′-4vk′+vk-1′2h}+v′k+1{3uk+1′-4uk′+uk-1′2h}]-H2a12[uk-1′{-3vk-1′+4vk′-vk+1′2h}+vk-1′{-3uk-1′+4uk′-uk+1′2h}],-3(vk+1-vk-1)+h(vk+1′+4vk′+vk-1′)=H2(2hgk′)-H2a11uk+1′{3uk+1′-4u′k+uk-1′2h}-H2a12uk-1′{-3uk-1′+4u′k-uk+1′2h},
whereak=a(rk),fk=f(rk),gk=g(rk),a11=ak+hak′+h22ak′′,a12=ak-hak′+h22ak′′,a10=ak.
The scheme (5.13) is free from the terms 1/k±1, hence very easily solved for k=1(1)N in the region (0,1). The system (5.13) can be solved using the NBSOR method.
Consider the boundary value problem [31]yiv-λyy′′=f(x),y(0)=V0,y(1)=V1,y′(0)=0,y′(1)=0.
This arises from the time-dependent Navier-Strokes equations for axisymmetric flow of an incompressible fluid contained between infinite disks that occupy the planes z=-d and z=d. The disks are porous and the fluid is injected or extracted normally with velocity V0atz=-d and V1 at z=d. Here, =d/ν, where νis the kinematic viscosity.
6. Numerical Illustrations
To illustrate our method and to demonstrate computationally its convergence, we have solved the following linear problem using the BGS method (See [25, 32–35]), whose exact solution is known to us. We have taken [0,1] as our region of integration. The right-hand side functions and the boundary conditions are obtained using the exact solution. The iterations were stopped when the absolute error tolerance became ≤10−12. All computations were performed using double length arithmetic.
Problem 1.
The problem is to solve (5.2) subject to the boundary conditions (5.4). The exact solution is u=r4sinr. The root mean square errors (RMSEs) are tabulated in Table 1. The graph of the errors for N=32 is given in Figure 1.
The RMSE.
2nd-order method
4th-order method
h
MAE
RMSE
MAE
RMSE
1/8
u
0.9512 (−03)
0.6737 (−03)
0.6018 (−03)
0.4089 (−03)
u′
0.3162 (−02)
0.2137 (−02)
0.2961 (−02)
0.1402 (−02)
1/16
u
0.2212 (−03)
0.1501 (−03)
0.3799 (−04)
0.2502 (−04)
u′
0.7021 (−03)
0.4995 (−03)
0.3022 (−03)
0.9846 (−04)
1/32
u
0.5410 (−04)
0.3576 (−04)
0.1989 (−05)
0.1304 (−05)
u′
0.1729 (−03)
0.1206 (−03)
0.2302 (−05)
0.5629 (−05)
1/64
u
0.1339 (−04)
0.8769 (−05)
0.9996 (−07)
0.6578 (−07)
u′
0.4298 (−04)
0.2972 (−04)
0.1562 (−05)
0.2983 (−06)
1/128
u
0.3363 (−05)
0.2174 (−05)
0.4198 (−08)
0.3345 (−08)
u′
0.1072 (−04)
0.7383 (−05)
0.9892 (−07)
0.1538 (−07)
Problem 2.
The boundary value problem is to solve (5.15). The exact solution is (1-x2)exp(x). The maximum absolute errors (MAE) and RMSE are tabulated in Table 2.
The MAE and RMSE.
2nd-order method
4th-order method
h
MAE
RMSE
MAE
RMSE
λ=1.0
1/8
u
0.8435 (−04)
0.5652 (−04)
0.2744 (−05)
0.1880 (−05)
u′
0.3552 (−03)
0.2322 (−03)
0.1058 (−04)
0.7178 (−05)
1/16
u
0.2292 (−04)
0.1499 (−04)
0.1854 (−06)
0.1217 (−06)
u′
0.7762 (−04)
0.5391 (−04)
0.6695 (−06)
0.4351 (−06)
1/32
u
0.5844 (−05)
0.3771 (−05)
0.1151 (−07)
0.7446 (−08)
u′
0.1885 (−04)
0.1322 (−04)
0.4115 (−07)
0.2618 (−07)
λ=100
1/8
u
0.1961 (−02)
0.1401 (−02)
0.5662 (−04)
0.3877 (−04)
u′
0.6514 (−02)
0.4687 (−02)
0.3824 (−03)
0.1910 (−03)
1/16
u
0.4808 (−03)
0.3312 (−03)
0.3821 (−05)
0.2558 (−05)
u′
0.1597 (−02)
0.1110 (−02)
0.1929 (−04)
0.9922 (−05)
1/32
u
0.1197 (−03)
0.8103 (−04)
0.2435 (−06)
0.1609 (−06)
u′
0.4023 (−03)
0.2714 (−03)
0.1135 (−05)
0.5843 (−06)
Problem 3.
The system of nonlinear equation (5.11) is to be solved subject to the natural boundary conditions. The exact solutions are u=cos(r) and v=exp(r). The MAE and RMSE are tabulated in Table 3.
The MAE and RMSE.
2nd-order method
4th-order method
h
MAE
RMSE
MAE
RMSE
1/10
u
0.7141 (−05)
0.4859 (−05)
0.2673 (−05)
0.1818 (−05)
v
0.3339 (−05)
0.2240 (−05)
0.6974 (−06)
0.4745 (−06)
u′
0.2488 (−04)
0.1683 (−04)
0.1266 (−04)
0.6538 (−05)
v′
0.1126 (−04)
0.8137 (−05)
0.2548 (−05)
0.1656 (−05)
1/20
u
0.1806 (−05)
0.1182 (−05)
0.2467 (−06)
0.1639 (−06)
v
0.8550 (−06)
0.5595 (−06)
0.4846 (−07)
0.3212 (−07)
u′
0.6486 (−05)
0.4128 (−05)
0.1167 (−05)
0.5946 (−06)
v′
0.2659 (−05)
0.1957 (−05)
0.1795 (−06)
0.1119 (−06)
1/40
u
0.4514 (−06)
0.2917 (−06)
0.2048 (−07)
0.1346 (−07)
v
0.2152 (−06)
0.1391 (−06)
0.3315 (−08)
0.2161 (−08)
u′
0.1620 (−05)
0.1020 (−05)
0.1027 (−06)
0.4911 (−07)
v′
0.6757 (−06)
0.4828 (−06)
0.1222 (−07)
0.7515 (−08)
1/80
u
0.1125 (−06)
0.7216 (−07)
0.1510 (−08)
0.9836 (−09)
v
0.5307 (−07)
0.3442 (−07)
0.2931 (−09)
0.1883 (−09)
u′
0.4038 (−06)
0.2525 (−06)
0.8190 (−08)
0.3662 (−08)
v′
0.1686 (−06)
0.1192 (−06)
0.9925 (−09)
0.6537 (−09)
1/160
u
0.3050 (−07)
0.1946 (−07)
0.1213 (−09)
0.7945 (−10)
v
0.1461 (−07)
0.9331 (−08)
0.1358 (−10)
0.8769 (−11)
u′
0.1076 (−06)
0.6812 (−07)
0.6420 (−09)
0.2913 (−09)
v′
0.4591 (−07)
0.3237 (−07)
0.5009 (−10)
0.3048 (−10)
Observe that for the linear singular problem for α=1, h1=1/64, h2=1/128, log(eh1/eh2)/log(h1/h2)=log(0.6578(-07)/0.3345(-08))/log2≈4. Thus, we obtain fourth-order convergence for u. Similarly, we get fourth-order convergence for u′.
7. Concluding Remarks
The numerical results confirm that the proposed finite difference methods yield second- and fourth-order convergence for the solution and its derivative for the fourth-order ordinary differential equation. Difference formulas for mesh points near the boundary are obtained without using the fictitious points. The proposed method is applicable to problems in polar coordinates and the derivative of the solution is obtained as the by-product of the method. We employ the BGS method to solve the block matrix systems of the linear singular problem and the BSOR method to solve the nonlinear singular problem. We have solved here a physical problem that arises in the axisymmetric flow of an incompressible fluid.
Acknowledgments
This research was supported by “The Council of Scientific and Industrial Research” under Research Grant no. 09/045(0836)2009-EMR-I. The authors thank the reviewers for their valuable suggestions, which substantially improved the standard of the paper.
TimoshenkoS. P.19612ndNew York, NY, USAMcGraw-Hill Bookxvi+5410134026AftabizadehA. R.Existence and uniqueness theorems for fourth-order boundary value problems1986116241542610.1016/S0022-247X(86)80006-3842808ZBL0634.34009AgarwalR. P.KrishnamoorthyP. R.Boundary value problems for nth order ordinary differential equations197972211230533788AgarwalR. P.Boundary value problems for higher order differential equations1981914761614645ZBL0467.34013AgarwalR. P.AkrivisG.Boundary value problems occurring in plate deflection theory19828314515410.1016/0771-050X(82)90035-3682885BaiZ.HuangB.GeW.The iterative solutions for some fourth-order p-Laplace equation boundary value problems200619181410.1016/j.aml.2004.10.0102189810GraefJ. R.KongL.A necessary and sufficient condition for existence of positive solutions of nonlinear boundary value problems200766112389241210.1016/j.na.2006.03.0282312596ZBL1119.34020GraefJ. R.QianC.YangB.A three point boundary value problem for nonlinear fourth order differential equations2003287121723310.1016/S0022-247X(03)00545-62010266ZBL1054.34038GreenbergM. D.2001chapter 7Engelwood Cliffs, NJ, USAPrentice HallMohantyR. K.A fourth-order finite difference method for the general one-dimensional nonlinear biharmonic problems of first kind2000114227529010.1016/S0377-0427(99)00202-21737078ZBL0963.65083NagleR. K.SaffE. B.1986chapter 6The Benjamin/CummingsppNoorM. A.Mohyud-DinS. T.An efficient method for fourth-order boundary value problems2007547-81101111110.1016/j.camwa.2006.12.0572398132ZBL1141.65375O'ReganD.Solvability of some fourth (and higher) order singular boundary value problems199116117811610.1016/0022-247X(91)90363-51127551ZBL0795.34018SchröderJ.Numerical error bounds for fourth order boundary value problems, simultaneous estimation of u(x)
and u″(x)198444223324510.1007/BF01410108753956ZBL0536.34004ShanthiV.RamanujamN.A numerical method for boundary value problems for singularly perturbed fourth-order ordinary differential equations20021292-326929410.1016/S0096-3003(01)00040-61905400ZBL1025.65044ZillD. G.CullenM. R.1997chapter 5NewYork, NY, USABrooks ColeUsmaniR. A.Finite difference methods for computing eigenvalues of fourth order boundary value problems19869113714310.1155/S0161171286000170837112ZBL0599.65057UsmaniR. A.SakaiM.Two new finite difference methods for computing eigenvalues of a fourth order linear boundary value problem198710352553010.1155/S0161171287000620896606ZBL0637.65081UsmaniR. A.TaylorP. J.Finite difference methods for solving [p(x)y″]′′+q(x)y=r(x)1983143-4277293727013ZBL0534.65040TwizellE. H.TirmiziS. I. A.Multiderivative methods for linear fourth order boundary value problems1984TR/06/84Department of Mathematics And Statistics, Brunel UniversityAgarwalR. P.ChowY. M.Iterative methods for a fourth order boundary value problem198410220321710.1016/0377-0427(84)90058-X739776ZBL0541.65055CabadaA.The method of lower and upper solutions for second, third, fourth, and higher order boundary value problems1994185230232010.1006/jmaa.1994.12501283059ZBL0807.34023FrancoD.O'ReganD.PeránJ.Fourth-order problems with nonlinear boundary conditions2005174231532710.1016/j.cam.2004.04.0132106442ZBL1068.34013HanG.LiF.Multiple solutions of some fourth-order boundary value problems200766112591260310.1016/j.na.2006.03.0422312608ZBL1126.34013HagemanL. A.YoungD. M.2004Mineola, NY, USADover Publicationsxx+3862096909MohantyR. K.EvansD. J.Block iterative methods for one-dimensional nonlinear biharmonic problems on a parallel computer199913323926310.1080/106371998089473691679647ZBL0922.68004ChawlaM. M.A fourth-order tridiagonal finite difference method for general non-linear two-point boundary value problems with mixed boundary conditions19782118393048878010.1093/imamat/21.1.83ZBL0385.65038MohantyR. K.EvansD. J.New algorithms for the numerical solution of one dimensional singular biharmonic problems of second kind199973110512410.1080/002071699088048831775821ZBL0942.65083EvansD. J.MohantyR. K.Alternating group explicit method for the numerical solution of non-linear singular two-point boundary value problems using a fourth order finite difference method200279101121113310.1080/002071602127041913846ZBL1003.65088PrescottJ.1961New York, NY, USADover Publications Incv+6660128114ElcratA. R.On the radial flow of a viscous fluid between porous disks197661191960412512ZBL0347.76022KellyC. T.1995Philadelphia, Pa, USASIAM PublicationSaadY.20032ndPhiladelphia, Pa, USASociety for Industrial and Applied Mathematicsxviii+5281990645VargaR. S.200027Berlin, GermanySpringerx+358Springer Series in Computational Mathematics1753713YoungD. M.2003Mineola, NY, USADover Publicationsxxvi+5702000714